IPv4

---

1. Historical Context and Design Philosophy

IPv4 was formalized in 1981 in RFC 791, during a time when networking was transitioning from isolated systems to interconnected networks.

It emerged from earlier protocols such as:

• NCP (Network Control Program) in ARPANET
• Early experimental internetworking protocols

The central problem IPv4 solves is:

How do we reliably deliver packets across multiple independent networks without requiring those networks to share internal structure?

This leads to the core design principle:

Best-Effort, Connectionless Delivery

IPv4 does not guarantee:

• Delivery
• Order
• Duplication avoidance

Instead, it provides:

• Logical addressing
• Packet forwarding

Reliability is delegated upward (e.g., TCP).

---

2. The Internet Layer in the TCP/IP Model

IPv4 operates at the Internet Layer of the TCP/IP model.

TCP/IP vs OSI Mapping

TCP/IP Layer	OSI Equivalent
Application	Application / Presentation / Session
Transport	Transport
Internet	Network
Network Access	Data Link + Physical

IPv4’s responsibilities:

• Logical addressing
• Routing
• Fragmentation
• Packet forwarding

---

3. IPv4 Addressing — Mathematical Foundation

An IPv4 address is a 32-bit unsigned integer.

This gives:

• Total possible addresses = 2³² = 4,294,967,296

Representation:

Decimal: 192.168.1.1
Binary:  11000000.10101000.00000001.00000001

---

4. Network vs Host — Formal Definition

An IPv4 address is divided into:

• Network prefix
• Host identifier

This division is defined by a prefix length (CIDR notation).

Example:

192.168.1.0/24

• First 24 bits → network
• Remaining 8 bits → hosts

Host Capacity Formula

Number of usable hosts:

2^h - 2

Where h = number of host bits

Why minus 2?

• Network address (all 0s)
• Broadcast address (all 1s)

---

5. CIDR — Classless Addressing

CIDR (RFC 4632) replaces inefficient classful addressing.

Key idea:

IP allocation is based on prefix length, not predefined classes.

Example:

10.0.0.0/13

• Flexible allocation
• Enables aggregation

Route Aggregation (Supernetting)

Multiple networks can be summarized:

192.168.0.0/24
192.168.1.0/24
→ 192.168.0.0/23

This reduces routing table size.

---

6. Subnetting — Network Engineering Tool

Subnetting divides a network into smaller logical segments.

Example

Original network:

192.168.1.0/24

Subnet into /26:

• Each subnet has 64 addresses
• 4 subnets total

Why Subnet?

• Reduce broadcast domains
• Improve performance
• Enhance security isolation

---

7. IPv4 Packet Structure (Deep Dive)

IPv4 transmits data in packets (datagrams).

Header Structure

Field	Size	Purpose
Version	4 bits	Always 4
IHL	4 bits	Header length
DSCP/ECN	8 bits	QoS
Total Length	16 bits	Packet size
Identification	16 bits	Fragmentation
Flags	3 bits	Fragment control
Fragment Offset	13 bits	Reassembly
TTL	8 bits	Loop prevention
Protocol	8 bits	Next layer
Header Checksum	16 bits	Error detection
Source IP	32 bits	Sender
Destination IP	32 bits	Receiver

---

8. Fragmentation and MTU

Different networks support different MTU (Maximum Transmission Unit).

If a packet exceeds MTU:

• It is fragmented

Key Fields:

• Identification
• Fragment Offset
• MF (More Fragments flag)

Problem:

Fragmentation is inefficient:

• Overhead
• Packet loss amplification

Modern networks try to avoid it using:

• Path MTU Discovery

---

9. Routing — How Packets Actually Move

Routing is based on longest prefix match.

Routers maintain a routing table:

Example:

Network	Next Hop
192.168.1.0/24	Router A
192.168.0.0/16	Router B

If destination = 192.168.1.5 → choose /24 (more specific)

---

10. ARP — Mapping IP to MAC

IPv4 uses ARP (Address Resolution Protocol) to map:

IP → MAC address

Process:

1. Broadcast ARP request
2. Target responds with MAC
3. Cached in ARP table

---

11. NAT — Extending IPv4 Life

Due to address exhaustion, NAT was introduced.

Types:

• Static NAT
• Dynamic NAT
• PAT (most used)

Key Concept:

Multiple private IPs → one public IP

Trade-offs:

Advantages:

• Conserves IP space
• Adds basic obfuscation

Disadvantages:

• Breaks end-to-end principle
• Complicates protocols (VoIP, P2P)

---

12. Special Address Spaces

Private Networks (RFC 1918)

• 10.0.0.0/8
• 172.16.0.0/12
• 192.168.0.0/16

Loopback

• 127.0.0.0/8

Link-Local (APIPA)

• 169.254.0.0/16

Multicast

• 224.0.0.0/4

---

13. Control Protocols Around IPv4

IPv4 does not work alone. It relies on supporting protocols:

• ICMP → error reporting
• ARP → address resolution
• DHCP → dynamic IP assignment

---

14. Scalability Problem — IPv4 Exhaustion

IPv4 exhaustion became critical around 2011.

Regional Internet Registries (RIRs):

• RIPE NCC
• ARIN

Solutions used:

• CIDR
• NAT
• Address reuse

---

15. Security Limitations

IPv4 lacks native:

• Encryption
• Authentication

Security is added via:

• IPSec
• Firewalls
• VPNs

---

16. IPv4 vs IPv6 — Engineering Perspective

Feature	IPv4	IPv6
Address size	32-bit	128-bit
Header complexity	Variable	Simplified
NAT	Required	Not required
Configuration	Manual/DHCP	SLAAC + DHCPv6

IPv6 solves:

• Address exhaustion
• Routing efficiency
• Built-in security

---

17. Why IPv4 Still Dominates

Despite limitations:

• Massive legacy infrastructure
• Cost of migration
• NAT effectiveness
• IPv6 adoption barriers

Most networks today are:

Dual-stack (IPv4 + IPv6)

---

18. Real-World Architecture Example

Typical enterprise network:

• Internal: Private IPv4
• Edge: NAT gateway
• External: Public IPv4
• Routing: BGP + internal routing (OSPF)

---

19. Key Engineering Insights

If you want to think like a network engineer:

• IPv4 is not just addressing — it’s a compromise system
• NAT is a workaround, not a solution
• CIDR is what keeps the internet scalable
• Routing efficiency is more critical than address count

---

20. Final Summary

IPv4 is:

• A connectionless, best-effort protocol
• Based on 32-bit logical addressing
• Extended by CIDR and NAT to remain viable
• Limited by address exhaustion and lack of built-in security

Yet, it remains:

The operational backbone of today’s internet.

Frequently Asked Questions

If IPv4 is connectionless and best-effort, why is it still reliable enough for everyday use?

IPv4 is reliable as a delivery mechanism, not as an end-to-end reliability system. It provides addressing and routing, while transport protocols like TCP add acknowledgments, retransmissions, and ordering when needed. This separation keeps the network layer simple and scalable, and lets applications choose the level of reliability they actually require.

Why does a /24 network have 254 usable hosts instead of 256?

A /24 contains 256 total addresses because 8 host bits remain. Two addresses are reserved: the all-zeros address identifies the network itself, and the all-ones address is the broadcast address. That leaves 254 assignable host addresses for devices inside that subnet.

What is the practical advantage of CIDR over the old classful system?

CIDR lets network blocks be sized according to actual need instead of fixed Class A, B, or C boundaries. That reduces wasted address space and makes route aggregation possible, so multiple nearby networks can be represented by a single summary route. The result is smaller routing tables and more efficient allocation.

How does longest prefix match work when multiple routes overlap?

Routers compare all matching routes for a destination and choose the one with the most specific prefix, meaning the longest subnet mask. For example, if both /16 and /24 match an IP, the /24 wins because it identifies a narrower range. This is what allows route summaries and specific exceptions to coexist.

Why is IPv4 fragmentation considered inefficient, and when does it become a problem?

Fragmentation adds overhead because each fragment needs its own header, and losing one fragment can force the whole original packet to be discarded. It also increases processing work for routers and hosts. It becomes especially problematic on paths with small MTUs, which is why Path MTU Discovery is preferred.

Why does IPv4 still need ARP if it already has IP addresses?

An IPv4 address identifies the logical destination, but Ethernet and similar link layers deliver frames using MAC addresses. ARP bridges that gap on a local network by resolving an IP address to the correct MAC address before transmission. Without ARP, a host would know where to send logically, but not physically.